Explanatory Notes Child and youth mortality trend series to 2019 United Nations Inter-agency Group for Child Mortality Estimation (UN IGME) Member agencies: UNICEF, the WHO, the UN Population Division and the World Bank Group September 2020 United Nations Explanatory Notes 2 The United Nations Inter-agency Group for relevant to the estimation of child and Child Mortality Estimation (UN IGME), which youth mortality including data from includes members from UNICEF, the World vital registration systems, population Health Organization (WHO), the United Nations censuses, household surveys and sample Population Division, and the World Bank Group, registration systems. was established in 2004 to advance the work on 2. Assess data quality, recalculate data monitoring progress towards the achievement of inputs and make adjustments if needed child survival goals. by applying standard methods. 3. Fit a statistical model to these data In accordance with the decision by the Statistical to generate a smooth trend curve Commission and the United Nations Economic that averages over possibly disparate and Social Council resolution 2006/6, UN estimates from the different data sources IGME child mortality estimates are produced for a country. in consultation with countries. UNICEF and the 4. Extrapolate the model to a target year, in WHO engage in joint country consultation on this case 2019. Sustainable Development Goal (SDG) indicators 3.2.1 (all countries aiming to reduce under-five To increase the transparency of the estimation mortality to at least as low as 25 deaths per process, the UN IGME has developed a Child 1,000 live births) and 3.2.2 (all countries aiming to Mortality Estimation (CME) web portal (www. reduced neonatal mortality to at least as low as childmortality.org). This portal shows country, 12 deaths per 1,000 live births), along with other regional and global estimates, includes all related child mortality indicators. available data on child mortality and indicates which data are currently officially used by UN The UN IGME released the new round of IGME. Once the new estimates are finalized, the estimates in September 2020. The estimates CME web portal is updated to reflect all newly will be also published in the United Nations SDG available data and the most recent estimates. Indicators Global Database, The State of the World’s Children by UNICEF and in the Global 2. Data Sources Health Observatory by WHO. Nationally representative estimates of mortality Methods used for the UN IGME child mortality can be derived from several different sources, estimates are summarized in this document, including civil registration and sample surveys. however procedures to estimate child, Demographic surveillance sites and hospital adolescent, and young adult mortality differ for data are excluded as they are rarely nationally Member States depending on data availability representative. The preferred source of data is a and type of data. These child mortality estimates civil registration system that records births and have been revised to take account of new data. deaths on a continuous basis. If registration is Therefore, this round of estimates may not be complete and the system functions efficiently, comparable with those published in the previous the resulting estimates will be accurate and 1 2 UN IGME reports or World Health Statistics . timely. However, in the developing world, most countries do not have well-functioning 1. Strategy vital registration systems, therefore household surveys, such as the UNICEF-supported Multiple UN IGME employs the following broad strategy Indicator Cluster Surveys (MICS) and the USAID- to arrive at annual estimates of child mortality: supported Demographic and Health Surveys (DHS), and periodic population censuses 1. Compile and assess the quality of all have become the primary sources of data on available nationally representative data childhood, adolescent and young adult (ages Explanatory Notes 3 0-24) mortality in developing countries. These For NMR, the number of deaths under one surveys ask women about the survival of their month of age and live births are used to calculate children and about the survival of their siblings, the neonatal mortality rate. which provide the basis for childhood, adolescent, and young adult mortality estimates for most In previous revisions UN IGME adjusted VR data developing countries. for incompleteness in the reporting of early infant deaths in several European countries. For more The first step in the process of arriving at details on the past adjustment see Notesi. estimates of levels and recent trends of childhood, adolescent and young adult mortality 2.1.2 Mortality among older children aged is to compile all newly available data and add 5─14 and youth aged 15─24 the data to the CME databases. Newly available data will include newly released vital statistics The calculation of the probability 10q5, the from a civil registration system, results from probability that a child aged 5 years dies before recent censuses and household surveys and, reaching his or her fifteenth birthday, is derived occasionally, results from older censuses or from a standard period abridged life table. The surveys not previously available. inputs are number of deaths for age group 5-9 years (noted D5-9) and for the age group 10-14 2.1 Data from civil registration systems years (D10-14), as well as the mid-year population for the same age groups (P and P ). 2.1.1 Under-five, infant mortality and neonatal 5-9 10-14 mortality • The death rate for age group 5-9, 5M5 is obtained by dividing by . For data from civil registration, the calculation of D5-9 P5-9 U5MR and IMR is derived from a standard period • The probability 5q5, which is the risk abridged life table. The inputs are number of of dying between age 5 and age 10, is obtained as q = (5 * M )/[1+(5- a ) * deaths for age group <1 year (noted D0) and for 5 5 5 5 5 5 M ], where a is the average number of the age group 1-4 years (D1-4), as well as the mid- 5 5 5 5 years lived by children who died in the age year population for the same age groups (P0 and group 5-9 (set at 2.5 for all countries) P1-4). • The same calculation is applied for 5q10. The formulae are as follows: • Finally, 10q5 = 1-(1-5q5)(1-5q10) Given that: The calculation of the probability q , the nqx is the probability of dying between age x 10 15 and age x+n, probability that an older adolescent aged 15 years th M = D /P , death rate for age <1, dies before reaching his or her 25 birthday, is 1 0 0 0 also derived from the number of deaths for the M = D /P , death rate for age group 1-4, 4 1 1-4 1-4 age groups 15-19 years (noted D ) and 20-24 Then: 15-19 years (D20-24), as well as the mid-year population q = M / [1+(1- a )* M ] 1 0 1 0 1 0 1 0 for the same age groups (P15-19 and P20-24), using where 1a0 is the fraction of year lived by the approach detailed above. an infant who died 1a0= 0.1 for low mortality country and In a few countries, vital registration data were 1a0= 0.3 for high mortality country incorporated to estimate mortality above age 5, despite being deemed too incomplete to be 5q0 = 1-(1-1q0)(1-4q1) where q = 4* M /[1+ (4- a ) * M ] used for under-five mortality. Civil registration 4 1 4 1 4 1 4 1 and vital statistics systems could capture a where 4a1 is the fraction of years lived by a child aged 1-4 years who died larger percentage of deaths of older children, a = 1.6 adolescents and young adults, as compared to 4 1 the deaths of young children, which are more Finally: IMR = 1q0 *1000 and U5MR= 5q0*1000 Explanatory Notes 4 likely to be unreported, especially when they FBH data, collected by all Demographic and occur in the neonatal period. Health Surveys (DHS) and increasingly also by Multiple Indicator Cluster Surveys (MICS), allow the calculation of child mortality indicators To select country-years for which vital registration for specific time periods in the past. DHS and data are included, and compute adjustment MICS usually publish under-five child mortality factors in case of incomplete registration, estimates for three 5-year periods before the we used a hybrid of the generalized growth survey, that is, 0 to 4, 5 to 9, and 10 to 14 years balance method (GGB) and the synthetic extinct before the survey.6,7,8 UN IGME has re-calculated generation method (SEG), the GGBSEG method, estimates for calendar year periods, using single which is one several demographic methods calendar years for periods shortly before the known as “death distribution methods”3 and survey, and then gradually increasing the number has been shown to perform better than the GGB of years for periods further in the past, whenever and SEG methods in isolation. The GGBSEG microdata from the survey are available. The method is implemented in the DDM package of cut-off points for a given survey for shifting from the R statistical software4. Completeness was estimates for single calendar years to two years, estimated for each country for periods between or two years to three, etc., are based on the pairs of recent censuses for which an age coefficients of variation (a measure of sampling distribution of the population was available in the uncertainty) of the estimates9. Demographic Yearbook5. When the estimated completeness was less than 80 per cent, mortality rates derived from vital registration In general, SBH data, collected by censuses and data were excluded from the model fit.